Matplotlib for Python Developers - Second Edition

Showing hierarchy in multivariate data with clustermap

Sometimes, a heatmap illustration can be hard to interpret when there are too many alternating color bands. This is because our data may not be ordered in terms of similarity. In this case, we need to group more similar data together in order to see the structure.

For this purpose, Seaborn offers the clustermap API, which is a combination of heatmap and dendrogram. A dendrogram is a tree-shaped graph that clusters more similar variables under the same branches/leaves. Drawing a dendrogram involves generally unsupervised hierarchical clustering, which is run in the background by default when we call the clustermap() function.

Besides unsupervised clustering, if we have a priori knowledge ...